AI Briefing#computer-vision
The latest AI and developer news about #computer-vision, with the original source and a short summary.
The latest AI and developer news about #computer-vision, with the original source and a short summary.
Four startups use NVIDIA hardware to automate sorting of plastics, recyclables, and textiles, diverting thousands of tons from landfills.
The new public benchmark found that general-purpose models significantly underperform specialized systems in detecting wrinkles and classifying skin types.
The system generates 9:16, 4:5, and 1:1 crops in a single pass using a three-stage pipeline of shot detection, focus region discovery, and temporal reframing.
The method achieves state-of-the-art quality on the Ava-256 benchmark while using 8x fewer Gaussians.
The new LoRA improves coordinate accuracy from 41% to 74% compared to the base model by returning page numbers and bounding boxes for every extracted field.
Krea has released Krea 3, a frontier image editing model claiming to be 10x faster and 3x cheaper than Nano Banana Pro, with a launch price of $0.06 per image.
The model uses a new dataset of 20,774 clips and a Motion Adapter to ensure view-consistent video generation.
The new AFP-GIC framework reduces decoder latency by 18.1% and parameter count by 20.5% compared to DC-VIC while supporting five bitrate operating points.
ORTUS AI has open-sourced RightWayUp, a 360-degree image rotation detection model under the Apache-2.0 license, which achieves 93.0% accuracy on held-out tests and demonstrates superior robustness against JPEG compression compared to existing models.
The open-source project replicates the 71-block Swin/ViT architecture of DLSS-NR build 310.8.0 using FP8 tensor cores, achieving byte-for-byte intermediate matches with the original.
RelateAnything supports an inference speed of 17.2ms per image on the RTX 3080 Laptop GPU using TensorRT FP32 and has released the RA-4M training corpus, which includes over 474,000 images and 4.3 million relationships.
Tencent has released WeVisDoc, a document parser based on Qwen3-VL, which ranked first on OmniDocBench.
SenseNova-Vision has been released, integrating detection, segmentation, and depth estimation into a single generative model without task-specific heads.
Apple researchers have released CapQuiz, a reference-free benchmark for evaluating the information fidelity of video captions.
The MultiMatte model, which fine-tunes SAM 3 with LoRA to isolate specific objects via text prompts, has been released, showing significant performance improvements on major benchmarks.
Three.js r186 integrates 3D Gaussian Splatting as a first-class engine feature, enabling realistic 3D rendering via WebGPU.
LG AI Research has unveiled an autonomous vision inspection framework that combines VLM and Agentic AI to eliminate human intervention from labeling to deployment.
A model fine-tuned on image deraining tasks using the depth estimation backbone and neck weights of YOLO26 has been released.
Flash models maximizing the efficiency of GigaPath and GigaTIME have been released, enhancing accessibility for large-scale pathology research.
The Tencent WeChat Vision team has released WeMM-Embedding, a general-purpose multimodal embedding model with 2B parameters that outperforms 8B baselines.
Apple released the Luce model for 3D asset generation and relighting.
Apple released STARFlow2, which combines LLMs and normalizing flows to generate text and images in a unified manner.
Apple proposed the IVT framework, which internalizes visual thinking during inference to reduce latency by more than 5x.
Roboflow has launched 'Roboflow Playground,' a tool that allows users to compare and test over 30 of the latest Computer Vision models side by side.